Merge pull request #94 from taburg/rpni-moore-initial-output-fix

Fix RPNI possibly discarding initial output of Moore machines

Author: Edi Muškardin

aalpy/learning_algs/deterministic_passive/rpni_helper_functions.py

@@ -60,6 +60,11 @@ def check_sequence(root_node, seq, automaton_type):
     Checks whether each sequence in the dataset is valid in the current automaton.
     """
     curr_node = root_node
+    # Check initial output for Moore machines and the like
+    if automaton_type != 'mealy' and len(seq) != 0:
+        if seq[0] is not None and seq[0] != root_node.output:
+            return False
+        seq = seq[1:]
     for i, o in seq:
         if automaton_type == 'mealy':
             if i not in curr_node.output or o is not None and curr_node.output[i] != o:
@@ -117,7 +122,7 @@ def _get_leaf_nodes(node):
     leaf_nodes = get_leaf_nodes(root_node)
     paths = []
     for node in leaf_nodes:
-        seq = []
+        seq = [] if automaton_type == 'mealy' else [root_node.output]
         curr_node = root_node
         for i in node.prefix:
             curr_node = curr_node.children[i]

tests/test_deterministic_passive.py

@@ -26,24 +26,21 @@ def prove_equivalence(self, learned_automaton):
 
         return correct_automaton == learned_automaton  # bisimilar
 
-    def generate_data(self, ground_truth, depth=5):
+    def generate_data(self, ground_truth, depth=5, step=1):
         data = []
         if isinstance(ground_truth, aalpy.automata.Dfa) or isinstance(ground_truth, aalpy.automata.MooreMachine):
             data.append(((), ground_truth.initial_state.output))
 
         alphabet = ground_truth.get_input_alphabet()
-        for level in range(1, depth + 1):
+        for level in range(1, depth + 1, step):
             for seq in product(alphabet, repeat=level):
                 ground_truth.reset_to_initial()
                 outputs = ground_truth.execute_sequence(ground_truth.initial_state, seq)
                 data.append((seq, outputs[-1]))
 
         return data
 
-    # TODO incomplete: skips input completeness options
     def test_all_configuration_combinations(self):
-
-
         automata_type = {Dfa: 'dfa', MooreMachine: 'moore', MealyMachine: 'mealy'}
         algorithms = ['gsm', 'classic']
 
@@ -68,3 +65,29 @@ def test_all_configuration_combinations(self):
 
         assert True
 
+    def test_all_configuration_combinations_input_incomplete_data(self):
+        automata_type = {Dfa: 'dfa', MooreMachine: 'moore', MealyMachine: 'mealy'}
+        algorithms = ['gsm', 'classic']
+
+        for automata in correct_automata:
+            correct_automaton = correct_automata[automata]
+            alphabet = correct_automaton.get_input_alphabet()
+            data = self.generate_data(correct_automaton, depth=3, step=2)
+            if automata_type[automata] == 'moore':
+                data += [(('a', 'a', 'a', 'a'), 1), (('b', 'b', 'b', 'b'), 2), (('c', 'c', 'c', 'c'), 3)]
+            for algorithm in algorithms:
+                learned_model = run_RPNI(data,
+                                         automaton_type=automata_type[automata],
+                                         algorithm=algorithm,
+                                         print_info=False)
+
+                is_eq = self.prove_equivalence(learned_model)
+                if not is_eq:
+                    print("Learned:")
+                    print(learned_model)
+                    print(algorithm, automata_type[automata])
+                    cex = compare_automata(learned_model, correct_automaton)
+                    print(cex)
+                    assert False
+
+        assert True